Jump to content

Release Notes (Geocaching® app: Favorite point indicator) - October 10, 2022


Recommended Posts

Release Notes (Geocaching® app: Favorite point indicator) - October 10, 2022


At Geocaching HQ, we're often testing and launching improvements and new features. This means you might see something in the Geocaching® app that your friend doesn't, or you could get a new feature or experience to try temporarily.

 

With today’s release, we are beginning such a test in the Geocaching® app.

 

Some of you will see a Favorite point indicator on the top 10% most Favorited caches on the map. At most, 300 caches load on the map, so approximately 30 caches will be highlighted. (In some cases, it will be more than 30, since we are including all caches that have the same amount of Favorite points as the cut off).

 

Why don’t we use FP percentage? How did we come up with the algorithm of top 10%?
We compared multiple different algorithms (Favorite points, Favorite point percentage, Wilson score, Fun score) and found that even though there were some differences, they were not significant within our samples (e.g. Favorite points and Wilson score resulted in a very similar set of highlighted caches). Favorite point percentage is more meaningful in some cases, but it also highlights newer caches that have few logs disproportionately. 

 

Knowing that no algorithm will ever be perfect, we decided to test Favorite points since they are already in the app and the algorithm is easy to explain and understand.

 

Nicole (nykkole), Product Manager, is watching this thread to answer questions whenever possible.

 

Any posts in this thread should relate to features in this release. Comments unrelated to the release may be removed. Please direct unrelated comments to other appropriate threads. Thanks!

  • Upvote 1
  • Funny 1
  • Helpful 1
Link to comment

A further observation on this. In my region I own 7 caches that are in the top 10% of FPs:

 

FP Rank     GC Code      Published       FPs      Total Finds     Finds this year

     14         GC6JMDK        2016             29                39                     0

     20         GC62WZJ        2015             22                29                     0

     32         GC5H5G2        2014             18                76                     3

     35         GC7YP51         2018             18                21                     1

     38         GC5YP8E         2015             17                27                     1

     47         GC5P0CE        2015              16               27                      2

     49         GC4ZQTF        2014              15               94                      0

 

What do you notice about these? They're all older caches which were reasonably popular in their day but now get few if any finds. Aren't these the very caches we were told early in the Year of the Hide that we should consider archiving?

  • Upvote 2
Link to comment

Discovered this new feature today while caching with another family member who uses the Official App. I think it's a great addition to the app, which allows you to quickly glance at the map and see what area you might want to cache in next.

I don't think I'll use to to plan my trips but is an excellent way to target a cache on the fly.

  • Upvote 1
  • Surprised 1
Link to comment

When I plan a tour, I always use Project-GC to show me the caches with the highest FP% or Wilson score. The effective number of favorite points is completely irrelevant! Regardless of this selection, you always have to look through the list manually anyway. How old/new is a cache, how old are the last finds or DNF and what do the logs say about it. There can be no correct automatic evaluation, but at most a recommendation. But I still say: The pure FP number says nothing.
 

  • Upvote 3
  • Helpful 3
  • Love 1
Link to comment

I saw this feature the other day and I like it.  I was initially confused though when I started playing with filters and noticed that some caches that had been flagged initially were no longer flagged.  Only when I read these RNs did I understand why.

 

I would like to see this feature have a configurable option that allows me to pick which algorithm I want used.  Maybe flag anything with > X FP.  Maybe anything > X% FP.  Maybe the top 10%.  Maybe something else.  But depending on what I'm looking for, I might want to choose a different algorithm.

 

Before anyone says that I can use filters to show anything with > X FP, I know that.  But that's not the objective here.  The objective is to flag caches with > X FP while still displaying everything else around the area.

 

Thanks for your consideration, GCHQ.  Great feature, could use some tweaking.  Keep up the good work.

  • Upvote 4
  • Helpful 2
Link to comment

While I think this is a great addition to the app (well done, very scannable!), I must agree with previous posters: the FP percentage and/or Wilson score is a way better indicator of cache greatness. Otherwise, especially as the total number of FPs awarded grows, the easy park 'n grabs with a funny cache container will always trump the great old multicache with only 50 finds.

Perhaps this could be a toggle in the app, as some cachers actually prefer easy funny containers.

  • Helpful 1
Link to comment

I'm not one of the test group it seems. I'm somewhat curious why this test wasn't conducted with the existing Experimental features function, where everyone with an interest could test it, and provide feedback.

 

Raw FP count is probably the right choice for the app. FP score is a fairly good indicator of accessibility of the cache, and that's worthwhile information when using the app. There's a place for multi-hour caching experiences, but doing those is not a choice I'd make when spontaneously looking at the app. When travelling, I want to find nice caches that I can do as a side activity, not caches that would take over my entire trip.

 

And of course the flagged caches change as I filter out everything I've already found, so even if the high FP% aren't in the first set, they will be flagged sooner or later.

  • Upvote 1
Link to comment
Just now, JL_HSTRE said:

 

If you're looking for caches in a smartphone live app you weren't going for a hike today anyway.

 

Most of the new cachers starting off in the game, around here at least, exclusively use the app and have never visited the website, so whether they're planning for a full-day hike or just a bunch of P&Gs, they're going to be basing their choice of caches on what the app shows.

  • Upvote 1
Link to comment
1 minute ago, barefootjeff said:

 

Most of the new cachers starting off in the game, around here at least, exclusively use the app and have never visited the website, so whether they're planning for a full-day hike or just a bunch of P&Gs, they're going to be basing their choice of caches on what the app shows.

 

The newbies who discovered geocaching through the smartphone app and plan their geocaching through the app without ever visiting the website are, by and large, not the kind of geocachers who go for full-day hikes, or even half-day hikes.

  • Upvote 2
Link to comment

What about rather than total Favorites or Favorite percentage, how about Favorite percentage among recent logs - either since 2011 (FPs were introduced in 2010)?

 

Personally, I find that Total Favorites is still useful as a threshold. Almost any cache with 10 Favorites is worth finding.

  • Funny 1
Link to comment
48 minutes ago, JL_HSTRE said:

Personally, I find that Total Favorites is still useful as a threshold. Almost any cache with 10 Favorites is worth finding.

 

Of the five caches I've placed in the last twelve months, only one (an urban Adventure Lab bonus) has reached 10 finders, let alone 10 FPs. The number of FPs is heavily biased in favour of older caches, particularly in areas where the number of cachers is in decline, and heavily biased in favour of tourist hotspots. One of the most highly favourited caches in my state, with 162 FPs, is a guard rail MKH overlooking the ferry wharf on Sydney harbour. It was published in 2004 and got those 162 FPs from 2421 finds. Does that make it the cream of the crop?

  • Helpful 1
Link to comment
10 hours ago, barefootjeff said:

One of the most highly favourited caches in my state, with 162 FPs, is a guard rail MKH overlooking the ferry wharf on Sydney harbour. It was published in 2004 and got those 162 FPs from 2421 finds. Does that make it the cream of the crop?

 

You're attacking a strawman. I said "Almost any cache with 10 Favorites is worth finding." I did not say caches with less than 10 aren't worth finding, nor anything about the best caches. 

 

Caches such as the ferry overlook you mentioned are a quandary to me. 

 

On the one hand, it sounds like it could have a really nice view.

 

On the other hand, it's a guardrail cache.

 

There may not be any better place to hide a container.

 

I found a scenic overlook cache in north Georgia (USA) that was a great view, but also a guardrail cache (pill bottle, so even worse than a magkey). It was a location I wouldn't have found without geocaching and a location I want to share with others, but a container that was "meh" at best. So does it deserve a Favorite?

 

It anything it's an example of how I think geocaching would be better with fewer containers and more Virtuals. But I've also seen ALs, which are psuedo-Virtuals, rapidly move from good locations towards lame locations, so other geocachers would manage to ruin that too. (Makes me glad the distribution of Virtual Rewards had been so limited.)

 

Edited by JL_HSTRE
  • Upvote 1
Link to comment
33 minutes ago, JL_HSTRE said:

It was a location I wouldn't have found without geocaching and a location I want to share with others, but a container that was "meh" at best. So does it deserve a Favorite?

I don't always follow my own principle, but to me location is the main point and the cache and container are secondary. So surely FPs should also be given based on location, even if the container isn't great.

 

But the cache page and container represent the COs effort more directly, so it feels bad to not give a FP for a gadget cache, even if the location is uninteresting. :rolleyes:

Link to comment
19 hours ago, mustakorppi said:
20 hours ago, JL_HSTRE said:

It was a location I wouldn't have found without geocaching and a location I want to share with others, but a container that was "meh" at best. So does it deserve a Favorite?

I don't always follow my own principle, but to me location is the main point and the cache and container are secondary. So surely FPs should also be given based on location, even if the container isn't great.

+1

 

As much as I dislike micros and as much as personal experience shows that statisticaly a micro delivers more offen a mundane experience than a small or regular, it is the experience *during* the hunt (which starts much earlier than immediately at GZ) which determines my enjoyment of a cache.

Link to comment

I'm missing an option to turn that feature off. It makes it even harder to see all the caches on the map and I usually don't care about FPs at all as I want to do them all anyways. So, for me it would be way more important to see the caches instead of the the hearts/numbers being put on top of some caches.

  • Upvote 3
  • Helpful 1
Link to comment
On 10/10/2022 at 10:55 PM, Geocaching HQ said:

Release Notes (Geocaching® app: Favorite point indicator) - October 10, 2022


Why don’t we use FP percentage? How did we come up with the algorithm of top 10%?
We compared multiple different algorithms (Favorite points, Favorite point percentage, Wilson score, Fun score) and found that even though there were some differences, they were not significant within our samples (e.g. Favorite points and Wilson score resulted in a very similar set of highlighted caches). Favorite point percentage is more meaningful in some cases, but it also highlights newer caches that have few logs disproportionately. 

 

 


I'd also second either percentage or Wilson as the pure number doesn't say much. My best example is "Lego" 10k+ FPs, but not a great cache at all. Percentage was comparably low and if the would be an option to downvote, I would been that on this cache for sure :-)

With just FPs that one would get promoted, while by far better might not. You're also arguing that new caches might be be promoted more. And yes, might be true, but also for a reason! Many older caches with high(er) FP counts nowadays either are in bad shape or aren't that special any longer. Think about a bird house, the first one to be found was so cool. But now they are so common, you might never get them a FP any more. That should be seen both at percentage and Wilson numbers that will go down over time, if a cache really is aging.

That would way more reflect the actual status of caches that are worth it, I'd say.

  • Helpful 2
  • Love 1
Link to comment

I choose the caches that I search for primarily based on location: I prefer parks and preserves.  Or, if I'm in a new state, I'll pick one that I think I can readily find because I may not have a lot of free time while visiting.  I do not even consider FPs.  I'm in the test group, and the little hearts are just noise.  If implemented, please provide a way to turn them off.

 

  • Upvote 1
  • Helpful 2
Link to comment

I agree with barefootjeff and others who have noted that absolute number of favorite points is less informative than favorite point % and that it results in a systematic bias towards older and easy to find caches.  In my experience, a cache with 20 finds and 8 fav points is going to be more interesting than a cache with 350 finds and 20 fav points.  If there is really a large overlap in the fav point % and total number of fav pts, then simply write the formula to compute it in two steps:  Fav Pt > 2; Fav Pt % > 25% would do it.  The first step eliminates 55% of all caches, the second highlights the small group of caches that are actually considered much better than average by folks who find them.  

edexter

  • Upvote 1
  • Surprised 1
  • Love 1
Link to comment
14 minutes ago, streudelz1222 said:

Could someone please help me understand what a Wilson Score or a Fun Score is?

The Wilson score is a measure based on the Percentage of Favorites, but corrected for the number of finds. In short, 90 Fav out of 100 Finds gives a better score than 9 out of 10. It works by determining the statistical bandwidth of the mean (90% in the example), then taking the lower bound of it. Or: the score for we’re at least 95% confident that represents the ‘real’ value.

For more maths, see https://www.evanmiller.org/how-not-to-sort-by-average-rating.html .

Link to comment

I would like to see a way to filter based on all of those measures. Total favorites can be good for finding the must-see but relatively easy to access spots. Favorite point percentage can be good for caches that may not be so easy to access or find. Wilson would be a good combination of both - some of the most awesome caches I've found didn't lead in total favorites, but had a really high Wilson score. No idea what the "fun score" is - sounds like something the hamsters made up :D

Link to comment

I like the favorites being displayed on the map. Various decisions are made to determine the caches that I will seek. Favorite points is not the sole criteria. If a cache is getting a lot of FP’s my curiosity is piqued and I want to see what is special about it. There are lonely caches that are harder to get to and require more effort that may not have many FP but will still be on my list. I cached with a friend today and we shared maps and he was disappointed that he does not have the FP’s displayed. I think you should allow all users to experience this feature and provide the option to turn off.

Link to comment
Guest
This topic is now closed to further replies.
×
×
  • Create New...